moved evaluation metrics to separate file

Author: Rpan4853

zephyr_ml/core.py

@@ -21,8 +21,8 @@
     "sklearn.metrics.precision_score",
     "sklearn.metrics.f1_score",
     "sklearn.metrics.recall_score",
-    "zephyr_ml.primitives.postprocessing.confusion_matrix",
-    "zephyr_ml.primitives.postprocessing.roc_auc_score_and_curve",
+    "zephyr_ml.primitives.evaluation.confusion_matrix",
+    "zephyr_ml.primitives.evaluation.roc_auc_score_and_curve",
 ]
 
 LOGGER = logging.getLogger(__name__)

zephyr_ml/primitives/evaluation.py

@@ -0,0 +1,64 @@
+"""
+Evaluation metrics
+"""
+
+import matplotlib.pyplot as plt
+import seaborn as sns
+from sklearn import metrics
+
+
+def confusion_matrix(
+        y_true,
+        y_pred,
+        labels=None,
+        sample_weight=None,
+        normalize=None):
+    conf_matrix = metrics.confusion_matrix(
+        y_true, y_pred, labels=labels, sample_weight=sample_weight, normalize=normalize
+    )
+    fig = plt.figure()
+    ax = fig.add_axes(sns.heatmap(conf_matrix, annot=True, cmap="Blues"))
+
+    ax.set_title("Confusion Matrix\n")
+    ax.set_xlabel("\nPredicted Values")
+    ax.set_ylabel("Actual Values")
+
+    ax.xaxis.set_ticklabels(["False", "True"])
+    ax.yaxis.set_ticklabels(["False", "True"])
+
+    return conf_matrix, fig
+
+
+def roc_auc_score_and_curve(
+    y_true, y_proba, pos_label=None, sample_weight=None, drop_intermediate=True
+):
+    if y_proba.ndim > 1:
+        y_proba = y_proba[:, 1]
+    fpr, tpr, _ = metrics.roc_curve(
+        y_true,
+        y_proba,
+        pos_label=pos_label,
+        sample_weight=sample_weight,
+        drop_intermediate=drop_intermediate,
+    )
+    ns_probs = [0 for _ in range(len(y_true))]
+    ns_fpr, ns_tpr, _ = metrics.roc_curve(
+        y_true,
+        ns_probs,
+        pos_label=pos_label,
+        sample_weight=sample_weight,
+        drop_intermediate=drop_intermediate,
+    )
+
+    auc = metrics.roc_auc_score(y_true, y_proba)
+    fig, ax = plt.subplots(1, 1)
+
+    ax.plot(fpr, tpr, "ro")
+    ax.plot(fpr, tpr)
+    ax.plot(ns_fpr, ns_tpr, linestyle="--", color="green")
+
+    ax.set_ylabel("True Positive Rate")
+    ax.set_xlabel("False Positive Rate")
+    ax.set_title("AUC: %.3f" % auc)
+
+    return auc, fig

zephyr_ml/primitives/jsons/zephyr_ml.primitives.postprocessing.confusion_matrix.json renamed to zephyr_ml/primitives/jsons/zephyr_ml.primitives.evaluation.confusion_matrix.json

@@ -1,12 +1,12 @@
 {
-  "name": "zephyr_ml.primitives.postprocessing.confusion_matrix",
+  "name": "zephyr_ml.primitives.evaluation.confusion_matrix",
   "contributors": ["Raymond Pan <[email protected]>"],
   "description": "Create and plot confusion matrix.",
   "classifiers": {
     "type": "helper"
   },
   "modalities": [],
-  "primitive": "zephyr_ml.primitives.postprocessing.confusion_matrix",
+  "primitive": "zephyr_ml.primitives.evaluation.confusion_matrix",
   "produce": {
     "args": [
       {

zephyr_ml/primitives/jsons/zephyr_ml.primitives.postprocessing.roc_auc_score_and_curve.json renamed to zephyr_ml/primitives/jsons/zephyr_ml.primitives.evaluation.roc_auc_score_and_curve.json

@@ -1,12 +1,12 @@
 {
-  "name": "zephyr_ml.primitives.postprocessing.roc_auc_score_and_curve",
+  "name": "zephyr_ml.primitives.evaluation.roc_auc_score_and_curve",
   "contributors": ["Raymond Pan <[email protected]>"],
   "description": "Calculate ROC AUC score and plot curve.",
   "classifiers": {
     "type": "helper"
   },
   "modalities": [],
-  "primitive": "zephyr_ml.primitives.postprocessing.roc_auc_score_and_curve",
+  "primitive": "zephyr_ml.primitives.evaluation.roc_auc_score_and_curve",
   "produce": {
     "args": [
       {

zephyr_ml/primitives/postprocessing.py

@@ -4,14 +4,12 @@
 
 import logging
 
-import matplotlib.pyplot as plt
 import numpy as np
-import seaborn as sns
 import sklearn
-from sklearn import metrics
 
 LOGGER = logging.getLogger(__name__)
 
+
 METRICS = {
     "accuracy": sklearn.metrics.accuracy_score,
     "precision": sklearn.metrics.precision_score,
@@ -85,59 +83,3 @@ def apply_threshold(self, y_proba):
         binary = [1 if x else 0 for x in y_proba > self._threshold]
         return binary, self._threshold, self._scores
 
-
-def confusion_matrix(
-        y_true,
-        y_pred,
-        labels=None,
-        sample_weight=None,
-        normalize=None):
-    conf_matrix = metrics.confusion_matrix(
-        y_true, y_pred, labels=labels, sample_weight=sample_weight, normalize=normalize
-    )
-    fig = plt.figure()
-    ax = fig.add_axes(sns.heatmap(conf_matrix, annot=True, cmap="Blues"))
-
-    ax.set_title("Confusion Matrix\n")
-    ax.set_xlabel("\nPredicted Values")
-    ax.set_ylabel("Actual Values")
-
-    ax.xaxis.set_ticklabels(["False", "True"])
-    ax.yaxis.set_ticklabels(["False", "True"])
-
-    return conf_matrix, fig
-
-
-def roc_auc_score_and_curve(
-    y_true, y_proba, pos_label=None, sample_weight=None, drop_intermediate=True
-):
-    if y_proba.ndim > 1:
-        y_proba = y_proba[:, 1]
-    fpr, tpr, _ = metrics.roc_curve(
-        y_true,
-        y_proba,
-        pos_label=pos_label,
-        sample_weight=sample_weight,
-        drop_intermediate=drop_intermediate,
-    )
-    ns_probs = [0 for _ in range(len(y_true))]
-    ns_fpr, ns_tpr, _ = metrics.roc_curve(
-        y_true,
-        ns_probs,
-        pos_label=pos_label,
-        sample_weight=sample_weight,
-        drop_intermediate=drop_intermediate,
-    )
-
-    auc = metrics.roc_auc_score(y_true, y_proba)
-    fig, ax = plt.subplots(1, 1)
-
-    ax.plot(fpr, tpr, "ro")
-    ax.plot(fpr, tpr)
-    ax.plot(ns_fpr, ns_tpr, linestyle="--", color="green")
-
-    ax.set_ylabel("True Positive Rate")
-    ax.set_xlabel("False Positive Rate")
-    ax.set_title("AUC: %.3f" % auc)
-
-    return auc, fig